FCT: DEE - Dissertações de Mestrado
URI permanente para esta coleção:
Navegar
Entradas recentes
- A importância do Cooling Effect no aumento da produção fotovoltaicaPublication . Tavares, Pedro Ferreira Rodrigues Catarino; Pronto, Anabela; Albuquerque, PedroEsta dissertação tem como principal objetivo a determinação do impacto que a água tem como método de arrefecimento das células dos painéis fotovoltaicos, através do estudo do caso da central fotovoltaica flutuante construída no Alqueva pela EDP em 2022. Primeiramente vai ser estudado o recurso fotovoltaico, a origem do mesmo e o seu funcionamento. De seguida são identificados outros métodos de arrefecimento usados em painéis fotovoltaicos, também são analisados os benefícios e desvantagens associados com estes, dando enfase ao método de arrefecimento usando água. Numa terceira fase do projeto é definida metodologia a usar para as simulações criadas, com o objetivo de compreender o verdadeiro impacto da água na produção final de energia. Posteriormente, é analisado o caso de estudo da central fotovoltaica flutuante no Alqueva e são descritas as condições em que os painéis solares se encontram. Após a definição das condições a observar em cada uma das simulações procede-se á análise dos resultados obtidos nas mesmas. Uma vez recolhidos e analisados individualmente, os resultados são comparados de forma a entender o impacto de parâmetros como inclinação, orientação e temperatura das células na produção final dos painéis solares. Por fim é discutida a importância que a água realmente tem como método de arrefecimento, não só no caso de estudo da central fotovoltaica flutuante do Alqueva como também em todas as centrais fotovoltaicas flutuantes.
- Solução Microgrid Aplicada a Zonas RemotasPublication . Cunha, Naylton Omair Vera Cruz; Pereira, PedroThis dissertation analyzes the application of microgrid-based solutions for the sustainable electrification of remote communities, focusing on technical, socioeconomic, and environmental feasibility. The work begins by identifying the energy needs of an isolated community in Sao Tome and Príncipe and proposes a hybrid system composed of photovoltaic generation and battery storage. Through simulations, different scenarios were studied, evaluating performance in terms of reliability, efficiency, and cost-effectiveness. The results indicate that the solar photovoltaic solution with battery storage, although requiring a very high initial investment, presents itself as the most viable alternative for realities like this, combining sustainability, emission reduction, and improvement in local quality of life, while minimizing dependence on fossil fuels. Beyond the technical component, the social and economic impacts are discussed, particularly in access to education, healthcare, food security, and microentrepreneurship opportunities. It is concluded that the implementation of microgrids constitutes a replicable strategy for other regions facing similar challenges, reinforcing the importance of public policy support and community involvement in the transition to inclusive and resilient energy models
- Integrative approach to analyze Water Use Efficiency in a Maize crop using Machine LearningPublication . Alves, Miguel Moura; Ferrada, Filipa; Correia, Patrícia; Silva, JorgeThe impact of global climate change on agriculture is apparent, with increasing drought scenarios leading to a reduction in crop yields. There is, therefore, an urgent need for new methods to improve growth in maize crops. In this context, it has been proposed the use of microbial biostimulants as a developing biological strategy to improve plant tolerance to water stress. This work encompasses a two-phased strategy: A first part consisting of an experimen- tal phase executed at "Asfertglobal" - a Portuguese company focused on the development of biostimulants and biocontrol solutions based on the use of microorganisms and new organic molecules - and a second phase, where machine learning algorithms are used to analyze and process the collected data, aiming to predict or estimate absolute values for various indexes. This approach provides additional information related to hydric stress and water-use efficiency (WUE), offering a large spectrum of parameters to support conclusions. This thesis focuses on applying machine learning algorithms to calculate water-use efficiency and predict the effects of different the biostimulants and how that effect is conditioned by the hydric regime; with the ultimate goal of analyzing and optimizing water consumption in maize plants. In parallel with this process, an interactive user interface was developed to streamline the company’s access to models and automation workflows. This interface accepts specific types of input and, in addition to applying the models to calculate water use efficiency, it automates processes based on received data. As a result, it functions not only as proof of a concept, but also as an industrial tool that will be used internally by the company.
- AUTOMATING MARKOV CHAINS: A SOFTWARE TOOL FOR TRANSITION PROBABILITY ESTIMATION, STATE MANAGEMENT AND FORECASTINGPublication . Fernandes, Ricardo José Fé; Sousa, Pedro; Guerreiro, GracindaMarkov chains are used in various fields, such as Physics, Computer Science, Economics and Finance, and the complexity of estimating the transition matrix of these mathematical models prompts the aim of this project. This dissertation proposes a methodology and a software tool to deal with building Markov chain models. When it involves large and/or complex datasets, the difficulty is essentially related to estimating transition probabilities, a task that can be difficult and time-consuming, especially as the number of states and/or the characteristics of the data set increase. The proposed solution is a tool that automates the process of calculating the transition matrix of discrete-time Markov chains. This tool, developed using React.js, allows users to define the states of the Markov chain, establish the transition rules between states and import data in panel format. This tool provides all of this through a user-friendly interface. By effectively handling large datasets and introducing dynamic cloud functionality, the tool can provide an effective solution to the difficulties associated with Markov chain modeling. We tested the tool on two datasets with different characteristics and the results were very promising, demonstrating its potential as a powerful tool for Markov chain modeling. With its ability to simplify complex transition rules and calculations as well as handle large datasets, this tool aims to make the modeling process faster, efficient and accessible.
- MODELING AND DESCRIBING THE OPERATOR’S ROLE AND INTERACTION USING ASSET ADMINISTRATION SHELLS IN INDUSTRY 5.0Publication . Martins, Miguel Filipe Simões; Rocha, André; Oliveira, FábioManufacturing is transitioning from an Industry 4.0 focus on automation and efficiency to Industry 5.0’s emphasis on human-centricity, resilience, and sustainability. The Asset Administration Shell (AAS) provides standardized digital representations of industrial assets, yet its systematic use for human operators remains insufficiently explored. This thesis investigates how operators can be modeled as first-class digital assets within the AAS. It proposes an isomorphic representation of humans and machines, comprising templates and instances, explicit training records, and a shared skills catalogue. This uniform semantics supports capability allocation, operator-in-the-loop execution, and traceability. The approach is validated in a smart-factory prototype that integrates AAS models authored with AASX Package Explorer and hosted on an AASX Server; orchestration via a JADE-based multi-agent system; a Unity HMI for operator interaction; and a Fischertech- nik/Arduino testbed for physical execution. Results show that embedding operators in the AAS enables consistent skill representa- tion, runtime coordination, and transparent provenance of actions across socio-technical resources. The contributions are: (i) a conceptual model for human-centric integration in the AAS, and (ii) an implemented prototype demonstrating feasibility and practical design guidance for sustainable, resilient factories aligned with Industry 5.0.
- Automated Detection of Contaminants: Using Computer Vision for Food Quality ControlPublication . Carvalho, Miguel Martins; Mora, André; Martinho, VergílioThe presence of foreign objects in food processing lines poses significant risks to consumer safety, product quality, and regulatory compliance. Traditional inspection methods such as manual visual checks or metal detectors are limited in their ability to detect non-metallic con- taminants like plastic, wood, or glass, especially when mixed with vegetable products. In re- sponse to this challenge, this work proposes the development of a real-time, computer vision system capable of detecting foreign materials on conveyor belts in a frozen vegetable factory. The proposed approach relies on RGB image acquisition and deep learning models, particu- larly object detection architectures such as YOLO (You Only Look Once), enhanced with at- tention mechanisms to improve small object recall. This document includes an overview of classical and modern vision techniques, the selection of suitable tools and models, and a defined work plan for implementation, calibration, testing and results. The expected outcome is a lightweight, accurate detection model integrated into a prototype system capable of operating in real-time under industrial conditions. This work aims to contribute to food safety automation in compliance with EU standards and to demon- strate the viability of deploying deep learning-based inspection systems on production lines.
- Solution to automatically create data connectors to support data exchanges between entitiesPublication . Horta, Gonçalo José Lopes; Gonçalves, RicardoNowadays, there is a constant need to stay connected to the world be it via social media, networking or work. The current global landscape practically forces all entities to be intertwined, facilitating knowledge sharing worldwide. On the enterprise level, most major companies have access to a wide variety of information regarding their products and clients that, by default are created according to each enterprise’s set of rules and structure. To help normalize this mismatch of structure, Digital Product Passports are used to share data among these companies, attaining interoperability and giving a more circular approach to the manufacturing economy. Not every company, however, has access to Digital Product Passports, making it a luxury rather than a standard. To help this need, this dissertation proposed a framework that will help small and medium enterprises to map the data sourced through files into a readable structure, so they are able to access the information coming from digital product passports and to send their own data through to them, achieving interoperability for themselves.
- Indicadores de Sustentabilidade Ambiental Hoteleira: Estudo Comparativo e Proposta de Indicador GlobalPublication . Martins, João Pedro Encarnação; Pina, JoãoO setor hoteleiro, pelas suas características operacionais e elevados padrões de conforto, apresenta um consumo energético intensivo, colocando desafios significativos à sustentabilidade ambiental. Neste contexto, o presente trabalho propõe uma metodologia de avaliação do desempenho energético de unidades hoteleiras, assente num conjunto de indicadores específicos e na construção de um índice composto — o Índice Global de Sustentabilidade Ambiental Hoteleira (IGSAH). Este indicador permite quantificar, de forma simples, direta e com reduzidos custos de implementação, o desempenho energético e hídrico global de cada hotel, através de uma abordagem estatística. A metodologia foi aplicada a uma amostra de hotéis, permitindo evidenciar tendências e identificar oportunidades de melhoria. Entre os principais desafios enfrentados destaca-se a escassez de contagens de energia e água desagregados por tipo de uso e de alguns dados operacionais da amostra de hotéis estudada o que limitou a quantificação precisa de alguns indicadores. Adicionalmente, a reduzida dimensão da amostra e a comparação entre hotéis de diferentes categorias constituem limitações à precisão dos resultados. Como trabalho futuro, sugere-se o refinamento do IGSAH com a integração de métricas de produção de energia renovável e de emissões de carbono, de modo a aumentar a sua representatividade ambiental. Propõe-se, para tal, a incorporação de seis indicadores: dois relacionadas com a geração renovável e os restantes quatro com as emissões de 𝐶𝑂2. Também se sugere a ampliação deste indicador às vertentes social e económica da sustentabilidade.
- Hardware Implementation of Advanced Encryption StandardPublication . Santos, Dinis de Jesus Oliveira Matias dos; Oliveira, Luís; Casaleiro, JoãoThis dissertation explores the design and evaluation of hardware implementations of the Advanced Encryption Standard (AES), following a structured progression from a software reference model to Register-Transfer Level (RTL), system integration and ASIC level analysis. The work begins with a software implementation used as a baseline and progresses to multiple Verilog RTL architectures. Two main architectures were evaluated and developed, iterative and fully pipelined, as well as alternative SubBytes implementations based on both lookup tables and composite field arithmetic. These two architectures are complete opposites, while one focus on minimising area the other aims to archive maximum throughput, these two design styles enable a deeper study of the relationship between performance and area usage. This work covers all three standard AES configurations (128, 192 and 256), with both encryption and decryption. This enables a broad comparison of architectural trade-offs. Finally, one last non-standard variation was explored, AES-512 (encryption). All designs were implemented using open-source tools and integrated into an FPGA system with an ESP32 that enabled an interface for end-to-end validation. All RTL cores were then synthesized and analysed with ASIC flows, allowing quantitative comparison in terms of area, timing and throughput. The final layout implementations provide physical insight and visual interpretation of the resulting digital circuits. Finally, the results obtained are quan- titatively compared with state of art AES hardware implementations.
- Aboveground, Belowground, Litter, and Shrub Biomass Estimation through Spaceborne LiDAR Remote SensingPublication . Silva, Diogo Marques da; Fonseca, José; Pereira-Pires, JoãoAccurate estimation of forest biomass is essential for understanding carbon sequestration dynamics and supporting climate change mitigation strategies. While aboveground biomass density has been extensively studied, other biomass pools, namely belowground biomass density, litter biomass density, and shrub biomass density, remain comparatively under-explored, despite their significant contribution to total carbon storage. This dissertation proposes and evaluates the use of machine learning regressors to estimate the four biomass pools jointly with spaceborne Light Detection and Ranging data from the GEDI and the ICESat-2 missions. Metrics derived from both missions, combined with ancillary features as predictors, were used in several regressions, including Random Forest, XGBoost, Multi-Layer Perceptron Regressor, Symbolic Regressor, and meta-ensemble techniques such as Voting Regressor and Stacking Regressor. Airborne Laser Scanning-derived biomass maps were used as reference data for training and validation. The model’s performance was evaluated using a bootstrapping technique and evalu- ation metrics (𝑅2, RMSE, MAE, Bias, and CV), along with qualitative analysis through scatter plots and histograms. The results demonstrate that the non-parametric models outperformed the parametric models, especially tree-based and ensemble regressors. The integration of ancillary features contributed to improving accuracy. The combination of GEDI and ICESat-2 coverage areas increased spatial density. The four biomass pools were estimated. The best GEDI models achieved rRMSE (RMSE) values of 29.37% (5.00 Mg/ha) for LBD in Spain, and 35.51% (6.04 Mg/ha) for SBD in Portugal. The best ICESat-2 models achieved rRMSE (RMSE) values of 23.55% (3.78 Mg/ha) for LBD in Spain, and 24.68% (3.19 Mg/ha) for LBD in Portugal. Finally, a QGIS plug-in prototype was developed to perform biomass estimation and generate GeoPackage outputs.